ChanServ changed the topic of #freedesktop to: https://www.freedesktop.org infrastructure and online services || for questions about freedesktop.org projects, please see each project's contact || for discussions about specifications, please use https://gitlab.freedesktop.org/xdg or xdg@lists.freedesktop.org
gnfzdz[m] has joined #freedesktop
co1umbarius has joined #freedesktop
columbarius has quit [Ping timeout: 480 seconds]
kusma has joined #freedesktop
scrumplex_ has joined #freedesktop
scrumplex has quit [Ping timeout: 480 seconds]
BLumia[m] has joined #freedesktop
tintou has joined #freedesktop
tinywrkb has joined #freedesktop
tomeu has joined #freedesktop
DavidHeidelberg[m] has joined #freedesktop
siddh has joined #freedesktop
ximion1 has quit [Quit: Detached from the Matrix]
sergi1 has joined #freedesktop
unrznbl[m] has joined #freedesktop
swick[m] has joined #freedesktop
zredshift[m] has joined #freedesktop
i-garrison has quit []
i-garrison has joined #freedesktop
AbleBacon has quit [Read error: Connection reset by peer]
gchini has quit [Quit: Leaving]
gchini has joined #freedesktop
sima has joined #freedesktop
jenatali has joined #freedesktop
bmodem has joined #freedesktop
seaweed[m] has joined #freedesktop
Hazematman has joined #freedesktop
todi has joined #freedesktop
dcunit3d has joined #freedesktop
dabrain34[m] has joined #freedesktop
alanc has quit [Remote host closed the connection]
alanc has joined #freedesktop
mripard has joined #freedesktop
damian has quit [Read error: Connection reset by peer]
damian has joined #freedesktop
tpalli has joined #freedesktop
<tpalli>
my ci pipeline is failing with "package `clap_builder v4.4.0` cannot be built because it requires rustc 1.70.0 or newer, while the currently active rustc version is 1.68.2"
<karolherbst>
daniels: that's not the rustc being used there
<karolherbst>
and I have no idea where that other comes from
<daniels>
hmm, I wonder if crosvm brings its own
<karolherbst>
well.. it has to as it's using 1.68.2, not 1.60
<karolherbst>
and I also don't see where else we install a rust toolchain, so dunno honestly
pixelcluster has quit [Ping timeout: 480 seconds]
pixelcluster has joined #freedesktop
havdan[m] has joined #freedesktop
atticf_ has joined #freedesktop
ximion has joined #freedesktop
atticf_ has quit []
atticf_ has joined #freedesktop
atticf has quit [Ping timeout: 480 seconds]
itaipu has joined #freedesktop
atticf_ has quit [Quit: leaving]
atticf has joined #freedesktop
bendlas[m] has joined #freedesktop
mripard has quit [Quit: mripard]
dabrain34[m]1 has joined #freedesktop
nirbheek_ has joined #freedesktop
atticf has quit [Quit: leaving]
ximion has quit [Quit: Detached from the Matrix]
atticf has joined #freedesktop
mripard has joined #freedesktop
bmodem has quit [Ping timeout: 480 seconds]
bmodem has joined #freedesktop
bmodem has quit [Remote host closed the connection]
bmodem has joined #freedesktop
bmodem has quit [Ping timeout: 480 seconds]
bmodem has joined #freedesktop
<DavidHeidelberg[m]>
daniels: Hi! What about the log size? Can you (I?) enable bigger logs?
<daniels>
ah yeah - I wonder if we're better storing those as artifacts, because as bentiss noted, we're currently carrying around like 70GB of job logs already in the database, which ... isn't ideal
bmodem has quit [Ping timeout: 480 seconds]
bmodem has joined #freedesktop
bmodem has quit [Ping timeout: 480 seconds]
bmodem has joined #freedesktop
<alatiera>
how long are they valid for
<alatiera>
after 1 month it's diminishing returns usually
bmodem has quit [Ping timeout: 480 seconds]
bmodem has joined #freedesktop
bmodem has quit [Excess Flood]
bmodem has joined #freedesktop
<daniels>
default validity is indeed 4 weeks
vkareh has joined #freedesktop
<DavidHeidelberg[m]>
I would love to extend just for some specific projects, but as I understand it it's runner configuration option
<daniels>
well, there's a gitlab-global limit, and then each individual runner can also expose a lower limit than that
<daniels>
but yeah, if you go back a few days, bentiss was mentioning that like 90% of our storage space is already taken up by build logs, so allowing them to be larger seems like not the right solution
<DavidHeidelberg[m]>
Right now, I have some jobs which just fails, but I won't get the output, I can redirect to the artifacts, but it doesn't seems right (also you won't get log before job finishes)
<DavidHeidelberg[m]>
it would fix my issue, but it doesn't feel like "right solution" from user viewpoint
<DavidHeidelberg[m]>
if I could set log expiration in the job, I would gladly set it for 7 days or even less in exchange for large log (ofc on other hand only for the jobs which needs large log)
itaipu has quit [Ping timeout: 480 seconds]
gkiagia has joined #freedesktop
<daniels>
I agree that would be a nice solution if it were possible, but right now there is no way to allow that just for certain projects
<daniels>
and tbf after looking at things like the rustc discussion above, having a longer-lived log of how images were built would be useful anwyay ...
razze[m] has joined #freedesktop
bmodem has quit [Ping timeout: 480 seconds]
AbleBacon has joined #freedesktop
<alatiera>
yes, we should be committing some info about how the images where built inside them
<alatiera>
the problem there isn't so much the log, but that we have no manifest like thing for docker tooling
itaipu has joined #freedesktop
Haaninjo has joined #freedesktop
Haaninjo has quit []
mripard has quit [Quit: mripard]
jarthur has joined #freedesktop
<bentiss>
yes. I finally pulled the registry data in equinix and setup the "new" registry that will only handle those. I also setup the fallback registry-gcs.fd.o for personal registries that were not migrated
<bentiss>
I was planning on making the change today, but it's a little bit too late. So I'll turn this one on on Sunday, and from then, the costs of the gcs should lower by a bit (but not this month as I pulled 6 TB yesterday)
<eric_engestrom>
btw, I don't remember, but is this one of the blockers for the 16.0 migration?
* eric_engestrom
is eager to use the new features :P
<bentiss>
eric_engestrom: this is mostly to reduce costs for the infra
<bentiss>
but yeah, once it's done, the next step is to migrate all the repos to the new cluster, then outage with the db migration, and then finally upgrade to 16.x
<bentiss>
eric_engestrom: but also that registry migration means that we can remove harbor and all the tricks we are doing to reduce costs on GCS
<bentiss>
so simplification of the infra
<eric_engestrom>
will this also reduce/remove the 500 issues with the registry?
<eric_engestrom>
all of this sounds really good btw, thanks for doing all that!
<bentiss>
eric_engestrom: hopefully yes
<eric_engestrom>
awesome :D
<eric_engestrom>
(sorry for all the questions, /me has clearly not been paying close enough attention ^^)
<bentiss>
right now when we push/pull a registry image we upload it/download it/re-push it
<bentiss>
so having a simple push will help IMO
systwi has joined #freedesktop
<bentiss>
eric_engestrom: no need to apologize, these are infra issues you can see mostly when you check on the google cloud bill :)
<eric_engestrom>
haha
* eric_engestrom
is very curious about all the internals, but also doesn't have the spare time to actually share the responsibility of being one of the admins
<mupuf>
bentiss: congrats! And thanks a lot for the communication! You rock!
<bentiss>
eric_engestrom: too bad, we would have gladly give you the keys...
<bentiss>
mupuf: thanks. Well, you'll probably be less happy next Monday when you'll have to copy the boot2container images...
<mupuf>
bentiss: they are in gfx-ci ;)
<mupuf>
We migrated!
<mupuf>
But the valve-infra ones will indeed need to be pushed
<bentiss>
well, anything created after last September will be on the new registry
<mupuf>
Then I just need to check when we last created the mesa trigger and telegraf containers
<bentiss>
well, it's the repository that matters
<mupuf>
The sync happened today, right?
<mupuf>
Well, anyway, i'll make myself available in Sunday to fix the fallouts
<bentiss>
so it's a little bit weird: if you have registry.fd.o/mupuf/foo/bar created before September, anything that you push here will not be on the new db
<bentiss>
but registry.fd.o/mupuf/foo/baz will if it was pushed after
<bentiss>
mupuf:thanks :)
<mupuf>
Oh, then I'll definitely need to re-push them
<mupuf>
No, thanks to you!
<alatiera>
you say september and I keep checking if it's still august
* alatiera
needs to go to sleep probably
<alatiera>
awesome stuff overall
<bentiss>
mupuf: you can cheat and check if this works -> in your /etc/hosts overwrite registry.freedesktop.org with one ip from c3-medium-dc.k8s.freedesktop.org, then skopeo inspect (with and without the overwrite)
<mupuf>
Oh, good idea. Will try that!
<bentiss>
or even simpler: if skopeo inspect docker://registry-gcs.freedekstop.org/mupuf/valve-infra/something:foo works, then there is a chance you'll have to re-push it
<bentiss>
(without the typos on the URL)
<mupuf>
Thanks, will do in a bit, when baby is asleep
sauce has joined #freedesktop
<bentiss>
mupuf: no rush. I'll go AFK probably until Sunday morning, so unless I'm pinged here and it's urgent, I'll probably leave that aside tomorrow
<daniels>
bentiss: I was going to say that I'm on holiday so should add a broadcast message that I'll not be around, but it looks like thunderstorms all week in les Landes, so I'll probably not be on the beach too much :P
<daniels>
bentiss: also _massively_ second mupuf's thanks for all the amazing behind-the-scenes work - we'd be dead without you!
heapify has joined #freedesktop
heapify has quit []
<bentiss>
daniels: oh nice, so you're going to France (after the heat wave then)
<daniels>
heh, yeah
<bentiss>
well, probably not the ideal week to be in France, but last week was hot
<bentiss>
like more than 40°C everyday (this week I mean)
<bentiss>
and today was better, only 37 :)
<daniels>
yeah, I mean there's more than one reason I left Australia, but that was one of them
todi has quit [Ping timeout: 480 seconds]
ximion has joined #freedesktop
todi has joined #freedesktop
flom84 has joined #freedesktop
ximion has quit [Quit: Detached from the Matrix]
flom84 has quit [Ping timeout: 480 seconds]
vkareh has quit [Quit: WeeChat 3.6]
Kayden has quit [Quit: wake up, nvme controller, it's not time to go to sleep]
Kayden has joined #freedesktop
psukys has joined #freedesktop
sima has quit [Ping timeout: 480 seconds]
Haaninjo has joined #freedesktop
psukys has quit [Ping timeout: 480 seconds]
<DavidHeidelberg[m]>
I think S3 handling overwriting files just fine, right?
<DavidHeidelberg[m]>
assuming the file haven't been used before and isn't cached anywhere